Multidimensional MarginalSparse throws with a simple example - "Random variables detected in the logp graph"

I am getting this error from the example, Sparse Approximations — PyMC example gallery is that expected too?